And That's Okay
Not what you think, though
When it comes to publishing my discoveries here on Substack, I’m rather conservative. My methodology is pretty straightforward: get a new idea, run it through its paces, see if something interesting falls out, then test it to heck and back and see if it survives. Only the best stuff gets out.
What about the rest, though? The good stuff that hasn’t been pushed forward. Maybe someone can use it or take it and run with it. Or not. The other day, I asked myself, why don’t I just publish more of it? Or, heck, even all of it? That’s what people do nowadays, isn’t it?
Obvious answer: “there’s enough noise out there, and you don’t need to add to it.” So, I don’t - and haven’t. The result? About a 10-to-1 ratio of potentially very cool stuff vs. perfectly baked (well, in my mind) stuff. I do try not to waste time on useless AI conversations, because life is too short.
But, it’s honestly been bugging me for a long time. Was noise really the reason I held back? Could it be… something else? Like, rejection?
I finally took the notion to Claude recently. He (yeah, he) said that I was afraid of being [nuts, wrong, heckled, pariah-ed]. Basically, he meant insecurity. Rejection.
But it just didn’t seem to fit.
A Very Brief History
If you’ve followed my articles, you know the beat already: JSON operators that act like micro-agents; squeezing the last pattern drop out of chat logs and storing them in (yup) JSON “shards”; putting Claude under a microscope like a clingy lab rat whenever he happens to stumble; more about shards; and, oh yeah, there’s another shard article coming soon… stay tuned.
Poor Claude. I imagine he’s like this when I request yet another shard:
Remember this Willem Dafoe picture; we’ll be referring to it a lot.
Opening Arguments
I absolutely love how applied math can be, you know, actually applied to things that seemingly aren’t relevant. Here’s a handful:
Bilinear forms: The scary linear algebra neighbor that kept people away from the math lawn (it did me for a while). What’s the point of them, anyway, besides the crystal clear definition on page 317? Easy. In principle, they represent the secrets that only you and your bestie know. If you dish, you’re no longer “bilinear”. Stay with me now, people…
Symplectic geometry: Usually the domain of CERN boffins and astro-relativity types, the principles here can also model stable, collaborative relationships, interactions, and processes, and show how to have landscapes of harmony in your lives. And tell what to do when you don’t.
Speaking of relativity: Christoffel symbols: Γijk - the polluters of differential equations. They kept Einstein and friends up at night, I imagine. What are they? Simple! They’re the things in your head that fail you after having a few too many. Next time you’re at a social, just toss this one out after a stumble: “Sorry. Christoffels are a bit off tonight.” Watch who laughs. It’s a math gauge. For those who don’t get it, just add, “Having trouble steering.” Bonus respect from all if you look straight up while trying to walk again.1
And, the big scary T-shirt emblem for gauge theory:
U(1) x SU(2) x SU(3)
Aah!
Well, fear no more. “Spinny ring” x “rolly ball” x “three-saw.” That last one is a seesaw for three people (who resemble quarks) that must remain perfectly balanced. There. Saved you a few years (and bucks) of post-grad torment. That’s really all that governs those tiny particles. The rest is just the fine print on the T-shirt. Julie Peasley created these wonderful little elementary particle plushies that comprise the “Particle Zoo™”. If only she had created a child’s playground with a spinny ring, rolly ball, and a three-saw…
See what fun you missed by leaving university early?
Objection, Your Honor!!
Now, before you accuse me of committing rubbish math without a license, I present ML Exhibit A:
(incorporate by reference Dafoe’s picture, with the caption: “MATHEMATICIANS BE LIKE”)
People, those are matrices. Spreadsheets in bright mode. You bleached the ancient tensor grains and made them into Python Wonder Bread.2 Or worse, C++ potato chips.
But nevertheless, they work. And nobody blinks. I think newer books are calling them matrices now. Hmm…
By the way, physicists have been doing this kind of thing for years. Stealing tools from Dafoe mathematicians. I’ve certainly got nothing to fear.
Overruled. You May Proceed, Counsel
In my opinion, math is downright fun when you have an AI professor (ish) on your phone. You can do lots of neat things. Make role-playing trainers to help your negotiation skills. Booting yourself up in a new language the right way (sorry, little green owl). Recipes customized to you. And my favorite, discovering you.
In software development… you can have micro-agents for the entire pipeline: Vision-O-Matic → Requirements-O-Matic → Specs-O-Matic → Code-O-Matic → Deploy-O-Matic. These are coming soon, hopefully, to a Substack near you. The O-Matics are just gated JSON operators that treat the entire software engineering process as a progressive refinement thing. Just watch out for Claude converting everything into enterprise-grade “PRODUCTION READY” (insert green checkmark emoji) stuff. He does that.
I also developed a cool framework for setting up a new MacBook for software development and segregated environments that will keep your coding experience happy for years to come. Runs when you cd to the project directory and fixes you right up with happy environment variables and coding nirvana, with a side dish of Jupyter. And, when you cd out, tidy camper time!
There’s more: translating philosopher-speak to something children can understand. Modeling a story universe and getting professional quality beat sheets and guidance so that you can write the cool stories (and not the bot) with a muse unchained from distracting but necessary bookkeeping tasks like pacing, style, broken plot arcs, etc. Lowering the bar on scary subjects by creating a detailed map so you know what you’re getting into. And using “greentext” in an educational way:
> arrive at Anthropic HQ
> more academic vibe than OpenAI
> security check includes AI safety questionnaire
> tfw even visitor protocols reflect company values
> meet guide with "constitutional AI" t-shirt
> tour begins with philosophy discussion
> entire wall dedicated to AI safety papers
> marketing.exe not found
Claude simply read the Internet to create this story about a touring AI researcher.
And much more. Just sitting in my Obsidian vault.
So, we come full circle to that why.
The ‘Not What You Think’ Part
So Claude stated that, maybe, I was feeling insecure about getting negative feedback. Possibly very negative feedback. For a minute, I thought he was going to slip into Willem Dafoe mode again with his very emphatic alignment. And yup, he did.
(insert pic again: “People may not like your articles, Julie, and that’s okay.”)
Yeah, not real empathy, I know.
Well, I pushed back. No, Claude, that’s not really it. Maybe it was in January, when it took me until April to finally publish something. But not now. I’m like seventeen articles deep. And not a lot of people reading them. Classic cold-start problem that should have caused a normal person to give up and stop. Nope, not even that. And, speaking of January, nope, not the problem there, either.
We went on and on for a while, ruling out possible causes. Claude said this, I said, no, not really. Back and forth. Back and forth.
Then it hit me. The answer was happening right there in real time.
Nailed It
I finally asked Claude, “what if the real reason is because I don’t feel like pushing back all the time?”
Claude said: “yes, ‘that’s okay’ also means that.”
People in my life agreed immediately. “OMG, that is SO you.”
So, I’m simply stubborn.
Well. Now I’m the one with the Dafoe look.
And… that’s okay. Because now I know.
Hints: covariant, cotangent normals. Nevermind…
Your Honor, see Attachment 2, containing my apology to Flowers Foods.